Agreement between pregnancies and pregnancy outcomes reported by husbands and wives was assessed in a sample of 857 couples interviewed between June 1989 and July 1990. The respondents were men employed in a semiconductor manufacturing plant in Burlington, Vermont, and their wives. The wives' reports were used as the standard against which the husbands' reproductive histories were evaluated. Measures included sensitivity, specificity, and percentage of agreement. Reports were considered to be congruent if an outcome reported by the husband agreed with the outcome reported by the wife within a period of +/- 6 months. Although men and women reported similar numbers of livebirths (1,478 and 1,500, respectively), men tended to misreport the timing of events; therefore, complete agreement on the numbers and dates of births was only 88.5%. Men also misreported the prevalence of low birth weight (sensitivity, 74%). Specificity was poorer for the younger (< 35 years) and less educated (< or = 12 years) respondents. Husbands' reports of spontaneous abortions had lower sensitivity (71.2%) than their reports of livebirths, particularly among the better educated (66.9%). Induced abortions were frequently omitted by the husbands (sensitivity, 35.1%), and events such as stillbirths or tubal pregnancies were too few in number to permit meaningful analysis. It is concluded that husbands' misreporting of their wives' reproductive histories may be substantial and sufficient to compromise the validity of epidemiologic studies. It would, therefore, be prudent to avoid the use of husbands as proxy informants of their wives' reproductive histories.
PIP: The reliability of husbands as proxy respondents for the provision of information on pregnancies and pregnancy outcomes for research on occupational hazards was assessed in a study conducted in 1989-90 in a semiconductor manufacturing plant in Burlington, Vermont (US). Full reproductive histories were obtained independently from 857 men 20-45 years of age who were employees of the plant and their wives. Husbands and wives volunteered information on 1833 and 1728 pregnancies, respectively, in the current marriage. Reports were considered to be congruent if an outcome reported by the husband agreed with one reported by the wife within a period of 6 months. Although husbands and wives reported similar numbers of live births (1478 and 1500, respectively), men tended to misreport the year of birth; complete agreement occurred in 88.5% of cases and agreement within 6 months in 92.2%. Specificity was poorer among men under 35 years of age and those with 12 years of education or less. Agreement on the timing of events was poor for other pregnancy outcomes as well. For example, with spontaneous abortion, complete agreement on the month and year of the event was only 31.7% and agreement within 6 months was 65.5%. For spontaneous abortion, sensitivity was 71.2% and specificity was 98.8% regardless of the time since the event. Sensitivity was lower among more educated respondents. Induced abortions were frequently omitted by husbands (sensitivity, 35.1%). These findings suggest that reliance on pregnancy histories derived from interviews with men may be problematic for epidemiologic research on male occupational exposures and reproductive outcomes.